Group B Streptococcal Beta-Hemolysin Expression Is Associated

INFECTION AND IMMUNITY, Sept. 1996, p. 3818–3826 Vol. 64, No. 90019-9567/96/$04.0010Copyright q 1996, American Society for Microbiology

Group B Streptococcal Beta-Hemolysin Expression IsAssociated with Injury of Lung Epithelial CellsVICTOR NIZET,1 RONALD L. GIBSON,2 EMIL Y. CHI,3 PAUL E. FRAMSON,1

MICHELLE HULSE,1 AND CRAIG E. RUBENS1*

Divisions of Infectious Disease1 and Neonatal and Respiratory Diseases,2 Department of Pediatrics,Children’s Hospital and Medical Center, Seattle, Washington 98105, and Department of Pathology,

University of Washington, Seattle, Washington 981953

Received 25 April 1996/Accepted 15 May 1996

Group B streptococci (GBS) are the leading cause of serious bacterial infection in newborns. Early-onsetdisease is heralded by pneumonia and lung injury, and the lung may serve as a portal of entry for GBS intothe bloodstream. To examine a potential role for GBS beta-hemolysin in lung epithelial injury, five wild-typestrains varying in beta-hemolysin expression were chosen, along with five nonhemolytic (NH) and five hyper-hemolytic (HH) variants of these strains derived by chemical or transposon mutagenesis. Monolayers of A549alveolar epithelial cells were exposed to log-phase GBS or stabilized hemolysin extracts of GBS cultures, andcellular injury was assessed by lactate dehydrogenase (LDH) release and trypan blue nuclear staining.Whereas NH strains produced no detectable injury beyond baseline (medium alone), hemolysin-producingstrains induced LDH release from A549 cells in direct correlation to their ability to lyse sheep erythrocytes. HHstrains were also associated with marked increases in trypan blue nuclear staining of A549 monolayers. Theextent of LDH release produced by HH strains was significantly reduced in the presence of dipalmitoylphosphatidylcholine, a known inhibitor of hemolysin and the major phospholipid component of humansurfactant. Electron microscopic studies of A549 cell monolayers exposed to HH GBS mutants revealed globalloss of microvillus architecture, disruption of cytoplasmic and nuclear membranes, and marked swelling of thecytoplasm and organelles. We conclude that GBS hemolysin expression correlates with lung epithelial cellinjury and may be important in the initial pathogenesis of early-onset disease, particularly when pulmonarysurfactant is deficient.

Group B streptococci (GBS) are the leading cause of seriousinfection in human newborns (4). Early-onset disease is her-alded by respiratory symptoms, caused by pneumonia withdense bacterial infiltration, alveolar hemorrhage, and inflam-matory exudate as characteristic histologic features (1, 19).Premature, low-birth-weight infants are at greatly increasedrisk for acquiring GBS pneumonia and invasive disease (9).The lung is a likely portal of entry for GBS into the blood-

stream, following aspiration of infected amniotic or vaginalfluids. Our laboratory has demonstrated that GBS are capableof invading alveolar epithelial (34) and endothelial cells (16),remaining within membrane-bound vacuoles. This process mayrepresent an initial step in the pathogenesis of invasive disease.In tissue culture, cellular invasion by GBS can occur at lowinocula without apparent damage to the host cells. It is there-fore likely that additional GBS virulence attributes play a rolein producing or eliciting the diffuse tissue injury seen in new-born pneumonia.The vast majority of GBS clinical isolates demonstrate beta-

hemolysis when plated on sheep blood agar (13). The degree ofbeta-hemolytic activity appears to correlate with the amount ofan orange carotenoid like-pigment produced by the organism(41). The GBS beta-hemolysin(s) itself has never been iso-lated, largely because it is unstable. High-molecular-weightcarrier molecules such as starch, albumin, or Tween 80 arerequired to preserve hemolytic activity in GBS culture super-natants (25). Interestingly, the hemolytic activity of such prep-

arations is inhibited by certain phospholipids (25) such as di-palmotyl phosphatidylcholine (DPPC), the major componentof human surfactant (31).More recently, stabilized extracts containing the GBS hemo-

lysin have been shown to alter the normal morphology ofMcCoy cells (mouse fibroblasts), suggesting the beta-hemoly-sin may affect a broader range of eukaryotic cell membranes(42). We sought to test the hypothesis that beta-hemolysinexpression might therefore contribute directly to injury of hu-man lung epithelial cells in vitro.In this report, we correlate the association of GBS beta-

hemolysin expression with injury of human alveolar epithelialcells in a tissue culture model. Chemical and transposon mu-tagenesis were used to produce nonhemolytic (NH) and hy-perhemolytic (HH) GBS variants, which were noninjuriousand hyperinjurious to epithelial cell membranes, respectively.

MATERIALS AND METHODS

Bacterial strains. Five clinical isolates of GBS were chosen for study: COH1,a highly encapsulated type III strain (47); COH31, a weakly encapsulated type IIIstrain (35); A909, a type Ia strain (24); UAB (also called Bib 501 Sm1), a type Ibstrain (17), and NCTC 10/84 (also called 1169-NT1), a type V strain (48). Allstrains were isolated from the blood of septic neonates except for COH31, whichwas isolated from a diabetic foot ulcer in an adult. Bacteria were maintained andgrown to mid-log phase in Todd-Hewitt broth (THB) to an optical density at 600nm of either 0.4 or 0.8 (equivalent to;108 or;109 CFU/ml, respectively) for usein hemolysin or cellular injury experiments.Chemical mutagenesis. NaNO2 mutagenesis of strain COH1 was performed

by using a modification of the technique described by Alper and Ames (2) in aprotocol biased to produce nucleotide deletions (36). Approximately 5 3 109

log-phase organisms in THB were pelleted by centrifugation and resuspended in300 ml of 0.1 M sodium acetate buffer (control) or 300 ml of the same buffercontaining 0.1 M NaNO2. The resuspended bacteria were incubated at 378C forvarious durations (between 1 and 20 min), and the reaction was stopped byadding 10 ml of THB on ice. The bacteria were once again pelleted by centrif-

* Corresponding author. Mailing address: Division of InfectiousDisease, Children’s Hospital and Medical Center, 4800 Sand PointWay N.E., Seattle, WA 98105. Phone: (206) 526-2116. Fax: (206)527-3890. Electronic mail address: [email protected].

3818

ugation, resuspended in 5 ml of fresh THB, and incubated at 378C for 1 h.Dilutions of these cultures were plated on Todd-Hewitt agar overnight, and theculture which resulted in 99% killing of COH1 compared with the control(10-min NaNO2 exposure) was selected for beta-hemolysin screening on trypticsoy agar–5% sheep blood agar.Transposon mutagenesis. Transposon Tn916 or Tn916DE mutagenesis of

GBS strains was performed as previously described, by conjugation with thehigh-frequency Enterococcus faecalis donor CG110 (47) or RH110 (32), respec-tively. Transposon Tn917 mutagenesis of strain A909 was accomplished by in-troduction of Tn917 on a conditional pWV01 family vector, pTV1OK, by elec-troporation. Transformants were amplified at the permissive temperature (308C)under erythromycin selection, and chromosomal transposition events were se-lected for by a shift to the nonpermissive temperature (398C) while maintainingerythromycin selection (15). Transposon mutants were screened for variations inhemolysin phenotype by plating on tryptic soy agar–5% sheep blood agar andovernight incubation at 378C.Southern blot analysis. Total cellular DNA was isolated from GBS by a

modification of the Hull protocol (21) in which 50 mg of mutanolysin per ml wassubstituted for lysozyme (23). DNA was digested to completion with the restric-tion enzyme EcoRI or HindIII, separated by 0.8% agarose gel electrophoresis,and transferred to nylon filters by the Southern method (38). Digoxigenin-labeled probes of (i) the Tn916DE-containing EcoRI fragment of plasmidpCER110 and (ii) the Tn917-containing EcoRI-PstI fragment of plasmidpTV1OK were prepared by using a Genius kit (Boehringer Mannheim). Theappropriate probe was hybridized to the target filter under standard conditions,and the probe-positive bands were visualized by chemiluminescence as recom-mended by the manufacturer.Phenotypic analysis of mutants. Chemical and transposon mutants were com-

pared with the wild-type parent strains for logarithmic growth in THB and RPMI1640 by optical density measurement. Production of GBS surface antigen wastested by latex agglutination (Streptex; Wellcome) and production of type-spe-cific capsule by immunoblot assay. Production of CAMP factor was assessed bystreaking each strain on Todd-Hewitt agar–5% sheep blood agar to within 3 mmof a perpendicular linear growth of Staphylococcus aureus. Acetoin production,the enzymatic activities of hippuricase, b-glucuronidase, alkaline phosphatase,leucine arylamidase, and arginine dehydrolase, and the ability to ferment 10sugar substrates were tested by using the API 20 Strep identification system forstreptococci (bioMerieux).Preparation of GBS hemolysin extracts. From logarithmic growth in THB,

1010 CFU of GBS was pelleted by centrifugation at 3,0003 g (Jouan GR412) for10 min, washed once in phosphate-buffered saline (PBS), and then resuspendedin 1 ml of PBS containing 0.2% glucose, 1% dextran, and 3% Tween 80 (25). Thesuspension was incubated at 378C in 5% CO2 for 30 min and centrifuged at3,0003 g for 10 min, and the supernatant (hemolysin extract) was removed fromthe bacterial pellet and stored on ice for not more than 1 h prior to use.Assay for hemolytic activity.Amodification of the method of Marchlewicz and

Duncan (25) was used to quantify GBS hemolysin activity, using whole bacteriaand the dextran-Tween 80 hemolysin extracts. From logarithmic growth in THB,108 CFU of GBS was pelleted by centrifugation at 3,000 3 g, washed once withPBS, and resuspended in 1 ml of PBS with 0.2% glucose. In a 96-well conical-bottom microtiter plate, 100 ml (107 CFU) of the bacterial resuspension wasplaced in the first well, and serial twofold dilutions in PBS-glucose were per-formed across the plate, each in a final volume of 100 ml. An equal volume of 1%sheep erythrocytes (RBC) in PBS-glucose was then added to each well, and theplate was incubated at 378C in 5% CO2 for 1 h. PBS-glucose alone and RBC lysiswith 0.1% sodium dodecyl sulfate (SDS) were used as negative and positivecontrols, respectively. After incubation, the plates were spun at 3,000 3 g for 10min to pellet unlysed RBC and bacteria, and 100 ml of the supernatant was

transferred to a replica plate. Hemoglobin release was assessed by measuringA420, and the hemolytic titer of a given strain was determined as the reciprocalof the greatest dilution producing 50% hemoglobin release compared with theSDS control. For determining the hemolytic titers of hemolysin extracts, 100 mlof extract was subjected to serial twofold dilutions in PBS-glucose and assayed inan identical fashion. The size of bacterial inoculum and duration of the assaywere standardized such that the weakly hemolytic wild-type strain COH1 and itshemolysin extract possessed a hemolytic titer of 1. All assays were performed induplicate and repeated three times.Epithelial cell cultures. A549 cells (American Type Culture Collection), a

human type II alveolar epithelial carcinoma cell line, were maintained andpassaged in RPMI 1640 tissue culture medium containing 10% fetal calf serum.Confluent monolayers of A549 cells in 96- or 24-well tissue culture plates (Corn-ing) or 8-well chamber slides (Nunc) were washed three times with PBS, andRPMI 1640 without fetal calf serum was added immediately prior to inoculationwith bacteria or hemolysin extracts in all experiments.LDH cytotoxicity assay. As in the assay for hemolytic activity, 108 CFU of

log-phase GBS was pelleted and washed but then was resuspended in 1 ml ofRPMI medium without fetal calf serum. A 100-ml aliquot (107 CFU) was addedto the first well of a 96-well culture plate containing a monolayer of A549 cells(;2 3 104 cells; multiplicity of infection of 500:1), and serial twofold dilutionswere performed in RPMI 1640 onto other monolayers across the plate. RPMI1640 alone and bacteria in RPMI 1640 without an A549 cell monolayer wereused as negative controls; distilled H2O lysis of a complete A549 cell monolayerwas used as a positive control. The plate was incubated at 378C in 5% CO2 for4 h, at which time a 20-ml aliquot of each supernatant was transferred to a replicaplate for lactate dehydrogenase (LDH) measurement using a miniaturized ver-sion of the Sigma colorimetric assay (catalog no. 500-C). To each well, 100 ml of0.1% NADH in standardized pyruvate substrate was added, and the plate wasincubated for 30 min at 378C. Sigma color reagent (100 ml) was then added toeach well for 20 min at room temperature. The A420 of each well was used tocalculate the residual pyruvic acid activity, which is inversely proportional toLDH activity. The cytotoxic titer of a bacterial strain was calculated as thereciprocal of the greatest dilution producing 50% LDH release compared withthe H2O lysis control. Cytotoxic titers for hemolysin extracts were determined ina similar fashion, and the duration of the LDH assay was again standardized suchthat COH1 and its extract possessed a cytotoxic titer of 1. All assays wereperformed in duplicate and repeated three times.Trypan blue dye exclusion assay. To monolayers of ;4 3 104 A549 cells in

eight-well chamber slides, 107 CFU of GBS resuspended in RPMI 1640 (multi-plicity of infection of 250) was added to a final well volume of 200 ml andincubated for 1 h at 378C in 5% CO2. After incubation, the supernatant wasremoved by gentle aspiration, 100 ml of 0.04% trypan blue in RPMI 1640 wasadded to each well, and the slides were returned to 378C for an additional 5 min.The supernatant was again removed, the monolayers washed once with PBS, andthe cells were fixed with 100 ml of 3% glutaraldehyde. Eosin (0.5%) was used as

FIG. 1. Southern blot analysis of EcoRI or HindIII chromosomal digests ofGBS type Ia strain A909, the NH Tn917 mutant A909-NH2, and the HH Tn917mutant A909-HH4. The probe is digoxigenin-labeled Tn917. As EcoRI does notcut within the transposon and HindIII cuts twice (liberating a 1.3-kb internalfragment), the pattern of one hybridizing band upon EcoRI digestion and threehybridizing bands upon HindIII digestion indicates that a single Tn917 insertioninto the A909 chromosome is associated with either the NH or the HH pheno-type.

TABLE 1. Isolation of GBS mutants with alterations inbeta-hemolysin phenotype

Phenotype Parent strain(capsule type) Mutagenesis Frequency

of mutants

Mutant chosen(no. of transposon

insertions)

NH COH1 (III) NaNO2 1 of 2,500 CM153Tn916DE 2 of 2,000 COH1-20 (1)

COH31 (III) Tn916 2 of 2,000 COH31c12 (2)A909 (Ia) Tn916DE 0 of 2,000

Tn917 3 of 3,000 A909-NH2 (1)UAB (Ib) Tn916DE —a UABH- (2)

HH COH1 (III) NaNO2 20 of 2,500 CM48Tn916DE 13 of 2,000 IN40 (1)

COH31 (III) Tn916 3 of 2,000 COH31c35 (1)A909 (Ia) Tn916DE 3 of 2,000 A909-HHA (2)

Tn917 4 of 3,000 A909-HH4 (1)

aMutant kindly provided by D. G. Pritchard, Birmingham, Ala.

VOL. 64, 1996 GBS HEMOLYSIN AND LUNG EPITHELIAL CELL INJURY 3819

a counterstain, and the monolayers were examined under high-power light mi-croscopy. The percentage of trypan blue-stained nuclei in three high-power fieldswas determined. Each sample was run in duplicate wells, and the assay wasrepeated twice.Electron microscopic studies. To monolayers of ;105 A549 cells in 24-well

tissue culture plates, 4 3 107 CFU of GBS resuspended in RPMI 1640 (multi-plicity of infection of 400) was added to a final well volume of 400 ml andincubated for 15 min at 378C. The supernatants were removed by gentle aspira-tion, and the monolayers were fixed with 3% glutaraldehyde and prepared fortransmission electron microscopy as described before (7).Inhibition of cellular injury by phospholipid. A549 cell monolayers in 24-well

tissue culture plates (;105 cells per well) were exposed to 107 CFU of HHwild-type strain NCTC 10/84 or HH Tn916DE mutant strain IN40 or to 100 ml ofhemolysin extract prepared from each strain. Experimental wells contained 500ml of RPMI or RPMI plus 100, 300, or 500 mg of DPPC (Sigma) per mlsolubilized by sonication for 1 min at 30 W. The monolayers were incubated at

378C with 5% CO2 for a total of 4 h. At hourly intervals, a 20-ml aliquot fromeach well was removed and assayed for LDH in the microtiter plate assaydescribed above. Monolayers incubated with RPMI 1640 at each concentrationof DPPC were used as negative controls; distilled H2O lysis of a complete A549cell monolayer was used as a positive control. Each condition was performed intriplicate and repeated two times. Data were expressed as percent LDH releasecompared with the positive control.

RESULTS

Selection of strains and isolation of mutants. To demon-strate that the association of beta-hemolysin production andepithelial cell injury is present across various GBS serotypes,five wild-type strains representing four capsular serotypes (Ia,

FIG. 2. Correlation of GBS beta-hemolysin expression with production of an orange pigment in GBS clinical isolates (top row), NH mutants (middle row), and HHmutants (bottom row) after overnight growth on tryptic soy agar–5% sheep blood agar (A) or overnight growth under anaerobic conditions on Todd-Hewitt agar (B).

TABLE 2. Hemolytic and cytolytic titers of GBS clinical isolates, chemical and transposon mutants with altered beta-hemolysin phenotype,and starch-Tween 80 hemolysin extracts from each strain

Strain Description

Titera

Hemolytic Cytolytic

Organism Extract Organism Extract

COH1 Type III clinical isolate 1 1 1 1CM153 NaNO2 mutant 0 0 0 0COH1-20 Tn916DE mutant 0 0 0 0CM48 NaNO2 mutant 32 32–64 64–128 128–256IN40 Tn916DE mutant 16–32 32 64 64–128

COH31 Type III clinical isolate 4 4 8–16 4–8COH31c12 Tn916 mutant 0 0 0 0COH31c35 Tn916 mutant 16 32 64–128 32–64

A909 Type Ia clinical isolate 4 4 8–16 4–8A909-NH2 Tn917 mutant 0 0 0 0A909-HH4 Tn917 mutant 16 16–32 32–64 64A909-HHA Tn916DE mutant 16 32 32–64 64–128

UAB Type Ib clinical isolate 8 4 16–32 8UABH- Tn916DE mutant 0 0 0 0

NCTC 10/84 Type V clinical isolate 32–64 32–64 512–1,024 128–256

a Inverse of the greatest dilution producing 50% lysis of sheep RBC (hemolytic titer) or 50% release of LDH from a monolayer of A549 alveolar epithelial cells(cytolytic titer), as determined in a microtiter plate assay. The initial inoculum in each assay was standardized such that the weakly hemolytic clinical isolate COH1possessed a titer of 1.

3820 NIZET ET AL. INFECT. IMMUN.

Ib, III, and V) were selected for study. Furthermore, to dem-onstrate that the association of altered beta-hemolysin produc-tion with changes in cytotoxicity was independent of themethod of mutagenesis, we studied NH and HH mutants cre-ated by using three different techniques: NaNO2 chemical mu-tagenesis, transposon Tn916 (Tn916DE) mutagenesis, andtransposon Tn917 mutagenesis. The frequencies of altered he-molysin phenotypes among the survivors of the NaNO2 mu-tagenesis protocol, transconjugates expressing the antibioticresistance marker of transposon Tn916 or Tn916DE, or Tn917transformants expressing erythromycin resistance are summa-rized in Table 1.Southern blot analysis of transposon mutants. To quantify

the number of transposon insertions, Southern blot analysiswas performed on chromosomal digests of each transposonmutant by using the appropriate transposon-specific digoxige-nin-labeled DNA probe. These studies revealed that transcon-jugate strains COH1-20 (NH) and IN40 (HH) contain onecopy of transposon Tn916DE, whereas strains A909-HHA(HH) and UABH-(NH) have two copies of Tn916DE (data notshown). NH transconjugate strain COH31c12 has two copies oftransposon Tn916 but has one insertion in common with anindependently isolated NH transconjugate, COH31c5 (South-ern analysis previously described [45]). HH strain COH31c35has a single copy of Tn916 (data not shown). Southern blotanalysis of Tn917mutants A909-NH2 and A909-HH2 reveals asingle band hybridizing to the Tn917 probe with EcoRI diges-tion (EcoRI does not cut Tn917) and three bands on HindIIIdigestion (HindIII cuts Tn917 twice, liberating a 1.3-kb inter-nal fragment), indicating that a single transposon insertion canbe associated with either an NH or an HH phenotype (Fig. 1).Two other independently isolated NH Tn917 mutants of strainA909 possessed the same EcoRI and HindIII hybridizationpattern as A909-NH2. Of four independently isolated HHTn917 mutants of strain A909, one double-insertion mutantshared the hybridizing bands of A909-HH4, whereas a distincthybridization pattern was found in the remaining two (data notshown). To summarize, in five of the eight transposon mutantsused in this study, a single chromosomal insertion of the trans-poson was associated with alteration of hemolysin phenotype.Phenotypic comparison of mutants with the wild type.

Chemical and transposon mutants with altered beta-hemolysinphenotype were tested for changes in other phenotypic traits to

look for possible pleiotropic mutations. All mutants expressedthe carbohydrate group B antigen, produced the correspond-ing type-specific capsular polysaccharide, and exhibited com-parable logarithmic growth in THB and RPMI 1640 comparedwith the wild-type parent strains (data not shown). In addition,all mutant strains possessed identical enzymatic profiles andsugar fermentation patterns on the API 20 Strep identificationsystem for streptococci as their respective parent strains. AllHH mutants, whether produced by chemical or by transposonmutagenesis, exhibited diminished, but not absent, CAMP fac-tor expression compared with parent strains. The HH wild-typestrain NCTC 10/84 also had significantly diminished CAMPfactor expression. NH mutants were identical to the parent inCAMP factor expression.Correlation of hemolytic activity and pigment production. A

close link between GBS expression of beta-hemolysin and pro-duction of an orange pigment has been suggested on bothepidemiological (28) and experimental (41) grounds. Figure 2compares the variation in beta-hemolysin activity among thewild-type and mutant strains after overnight growth on bloodagar plates with the variation in pigment production on Todd-Hewitt agar after overnight growth under anaerobic condi-tions, known to favor pigment production (27). Increased beta-hemolysin activity among the five wild-type strains wasassociated with increased expression of an orange pigment.NH mutants were associated with loss of pigment expression.HH mutants expressed increased pigment compared with theparent strains. The link between the degree of expression ofGBS beta-hemolysin and orange pigment was therefore main-tained following mutagenesis of the beta-hemolysin pheno-type.Quantification of hemolytic activities of wild-type GBS

strains and hemolysin mutants. A liquid-phase microtiter di-lution assay in PBS-glucose was used to quantify the hemolysinproduction of each wild-type strain and mutant. The hemolytictiters of the five wild-type GBS strains studied are shown inTable 2. Strains COH31 and A909 produced 4 times, strainUAB parent produced 8 times, and strain NCTC 10/84 pro-duced 32 to 64 times as much hemolytic activity in the micro-titer well assay as the weakly hemolytic strain COH1. Stan-dardized such that the COH1 titer equaled 1, the hemolytictiter of starch-Tween 80-stabilized extracts of each strain cor-responded to the hemolytic titer of the intact organism. NHmutants, whether created by chemical or by transposon mu-tagenesis, produced no detectable hemolytic activity in themicrotiter assay, nor was hemolytic activity extractable withstarch-Tween 80. HH mutants, whether created by chemical orby transposon mutagenesis, produced 4 to 32 times as muchhemolytic activity as the parent strain, an increase which waspreserved in starch-Tween 80 extracts. Alteration of beta-he-molysin expression by intact organisms was thus accompaniedby a consistent alteration in the ability of the hemolysin to bestabilized or extracted by the starch-Tween 80 carrier. PBS-glucose alone did not produce RBC lysis. The hemolytic titersof each strain or starch-Tween 80 extract were consistent andreproducible within 1 or 2 dilutions.LDH release from lung epithelial cells. To test the correla-

tion of GBS beta-hemolysin expression with epithelial cell in-jury at a biochemical level, we measured release of the eukary-otic cytoplasmic enzyme LDH from A549 alveolar epithelialcells. Injury to cell monolayers exposed to serial dilutions ofbacterial inocula was quantitated by LDH release in a micro-titer plate assay and is reported as a cytolytic titer standardizedsuch that the titer of wild-type strain COH1 equaled 1.Whereas NH strains produced no cellular injury beyond base-line (medium alone), hemolysin-producing strains induced

FIG. 3. Correlation of hemolytic titers of GBS clinical isolates and hemolysinmutants with trypan blue staining of A549 alveolar epithelial cell nuclei in tissueculture monolayers exposed to each strain. ■, wild-type strain; 1, NaNO2mutant; o, Tn916 mutant; p, Tn917 mutant.


3822

LDH release from A549 cells in direct correlation to theirability to lyse sheep RBC (Table 2). HH mutants producedequivalent LDH release at inocula 4- to 128-fold lower thanthose used for the parent strains from which they were derived.Cytolytic activity was stabilized in starch-Tween 80 extracts inproportion to the cytolytic titer of the intact organism. Thecytolytic titers of each strain or starch-Tween 80 extract wereconsistent and reproducible within 1 or 2 dilutions. It thereforeappears that the cytolytic activities of the GBS strains and theirrespective mutants tested correlated directly with their hemo-lytic activities and that the cytolytic factors were stabilized bythe same carrier molecules as the hemolytic factors.Trypan blue nuclear staining. The association of GBS he-

molysin expression with A549 pneumocyte injury was also as-sessed histologically by using the vital dye trypan blue. This dyestains the nuclei of cells which have lost the membrane pumpfunction to exclude it. Incubation of A549 cells with NH orweakly hemolytic strains was associated with little or no trypanblue staining, whereas incubation with HH mutants or the HHwild-type strain NCTC 10/84 resulted in positive staining of 35to 100% of nuclei (Fig. 3). Trypan blue staining would appearto be less sensitive for low-level epithelial cell injury than LDHrelease but corroborates the biochemical evidence of cytotox-icity produced by the HH strains.Electron microscopy. Electron microscopy was used to study

the ultrastructural changes associated with GBS hemolysin-induced epithelial cell injury. Figure 4 compares transmission

electron micrographs of A549 pneumocytes exposed for 15 minto identical inocula of the weakly hemolytic parent strainCOH1, an HH Tn916DE mutant (IN40), and an HH chemicalmutant (CM48). The pneumocyte exposed to strain COH1exhibits an (i) intact cellular membrane with microvilli and (ii)dense regular cytoplasmic contents, organelles and chromatin.Pneumocytes exposed to the HH mutants exhibited dramaticloss of cytoplasmic density, breaks in the cytoplasmic mem-brane, dilated mitochondria, intracytoplasmic vacuole forma-tion, frequent splitting of the nuclear membrane, and clumpingof nuclear chromatin. Figure 5 compares scanning electronmicrographs of A549 pneumocytes exposed to strain COH1and the HH mutant CM48. Exposure to the HH mutant resultsin loss of microvillus architecture and bleb formation on thecell surface.Inhibition by surfactant phospholipid. Phospholipid com-

ponents of human surfactant are known to inhibit GBS beta-hemolysin-associated RBC lysis (25) and alterations of McCoycell morphology (42). We sought to assess whether phospho-lipid might also be protective for lung epithelial cells exposedto high levels of GBS hemolysin. Using LDH release to mea-sure injury of A549 cell monolayers, we added increasing con-centrations of the phospholipid DPPC to the tissue culturemedia by sonication and exposed the monolayers to HH wild-type strain NCTC 10/84, HH transposon mutant IN40, andstarch-Tween 80 hemolysin extracts from these strains (Fig. 6).Medium alone was used to define baseline (0%) LDH release,

FIG. 4. Ultrastructural features of GBS hemolysin-associated epithelial cell injury. Transmission electron micrographs show A549 pneumocytes exposed to theweakly hemolytic parent strain COH1 (A), HH Tn916DE mutant IN40 (B), and HH chemical mutant CM48 (C). Arrows indicate surface bleb formation, dilatedmitochondria, and intracytoplasmic vacuole formation (B) and splitting of the cytoplasmic and nuclear membranes (C). Loss of cytoplasmic density and clumping ofnuclear chromatin are seen in panels B and C.


and H2O lysis of the monolayer defined 100% LDH release;LDH quantitation itself was not affected by the range of con-centrations of DPPC tested. A dose-related decrease in LDHrelease was found with increasing DPPC concentration follow-ing exposure of the epithelial cell monolayers to either intactHH GBS organisms or starch-Tween 80 hemolysin extracts. Inthe presence of 500 mg of DPPC per ml, there was .90%reduction in LDH release at 1 h of exposure in all cases. Thesedifferences were persistent over 4 h of exposure with thestarch-Tween 80 extracts, but growth of either strain of intactbacteria ultimately overcame the DPPC inhibition and pro-duced high-level injury by 4 h in both strains.

DISCUSSION

The studies described above demonstrate that the expres-sion of beta-hemolytic activity of GBS can be correlated di-rectly with injury of lung epithelial cells in vitro. Chemical andisogenic transposon mutants of GBS strains which are NH arenoninjurious, and those which are HH are hyperinjurious, tolung epithelia. The GBS product(s) associated with epithelial

cell injury is stabilized in culture extracts by the same com-pounds (starch-Tween 80) which stabilize activity againsterythrocyte membranes. Thus, it appears that the GBS beta-hemolysin(s) possesses a broader range of host cell specificityand may be more appropriately classified as a membrane-disrupting cytolysin. Several other gram-positive hemolysins,e.g., the alpha-toxin of S. aureus (43), streptolysin O of Strep-tococcus pyogenes (6), and the plasmid pAD1-encoded hemo-lysin-bacteriocin of E. faecalis (22), are known to injure mam-malian epithelial cells.The GBS beta-hemolysin has yet to be isolated. Some evi-

dence exists to suggest that it is normally attached to thebacterial surface membrane (30). Extracellular hemolytic ac-tivity is rapidly lost unless high-molecular-weight stabilizermolecules are present (25). Because the hemolysin does notalter the gel column elution behavior of various carrier mole-cules, it appears to be a small molecule (44). Sensitivity to theprotease subtilisin suggests that it is a protein (26), but at-tempts to visualize it by gel electrophoresis or to raise antibodyagainst crude preparations have been unsuccessful (11, 26).

FIG. 5. Additional ultrastructural features of GBS hemolysin-associated epithelial cell injury. Scanning electron micrographs show A549 pneumocytes exposed tostrain COH1 (A) and HH chemical mutant CM48 (B). Loss of microvillus architecture and bleb formation on the cell surface of the hyperhemolytic mutant are evident.


The cloning of a GBS genetic determinant in a pUC8 vectorwhich allowed Escherichia coli host strains to express beta-hemolysis on blood agar has been reported (10). An openreading frame with a deduced amino acid sequence of 230residues, including a possible hydrophobic signal sequence of19 amino acids at the N-terminal end, was identified. Thisputative GBS hemolysin gene had no homology with anyknown bacterial hemolysin or streptococcal protein.The electron microscopic studies of GBS beta-hemolysin-

induced injury to lung epithelial cells are suggestive of a pore-forming toxin. Discrete membrane disruptions, cellular swell-ing, loss of intracytoplasmic density, and changes in organellesand chromatin are consistent with entry of water into the celland hypo-osmotic damage, analogous to changes seen with S.aureus alpha-toxin (5) or the terminal membrane attack com-plex (C5b-9) of human complement (29). Earlier work hadshown that radiolabeled rubidium (86Rb1) and hemoglobindemonstrated identical efflux kinetics from sheep RBC follow-ing exposure to GBS beta-hemolysin (26), suggesting that thetoxin produces membrane lesions of large size.Injury of lung epithelial cells by GBS beta-hemolysin in vitro

suggests a possible pathogenic role for this molecule in humandisease. Direct damage to host cell membranes could contrib-ute to the severe pneumonia characteristic of early-onset GBSinfection. Disruption of the epithelial cell barrier might alsofacilitate access to the bloodstream and systemic spread by theorganism. In the microtiter LDH release assay, injury of lungepithelial cells is detected at bacterial inocula of between 106

GBS per ml (most hemolytic clinical isolate NCTC 10/84) and108 GBS per ml (least hemolytic clinical isolate COH1). WhenGBS pneumonia was induced in newborn primates, bacterialdensity reached 109 to 1011 organisms per g of lung tissue (33),indicating a large potential reservoir for beta-hemolysin pro-duction and epithelial cell injury.There is precedent for pore-forming bacterial hemolysin-

cytolysins as virulence factors involved in producing pneumo-

nia or lung injury. S. aureus alpha-toxin (37) and Escherichiacoli hemolysin (Hly) (12) produce thromboxane-mediated va-soconstriction and edema formation in isolated, perfused rab-bit lungs and may be implicated in the development of septiclung failure. More recently, two RTX family hemolysins of thegram-negative bacterium Actinobacillus pleuropneumoniaewere shown to be important virulence factors in the productionof hemorrhagic and necrotic lung infections in swine (39).Knowledge of the effects of GBS beta-hemolysin production

on virulence in animal models is limited. NH mutant strainCOH31 clone 12 did not show an increased 50% lethal dosecompared with the parent strain following subcutaneous injec-tion in neonatal rats (45). In contrast, intravenous administra-tion of partially purified GBS hemolysin extracts in rabbits orrats produced dose-dependent hypotensive changes and a lim-ited number of deaths due to shock, findings not produced bystreptolysin S from Streptococcus pyogenes (18). Both models,however, bypass the initial interaction with the lung epithelialbarrier critical to the pathogenesis of GBS pneumonia. Inperhaps a more relevant model, Wennerstrom et al. (46) usedNH and HH chemical mutants of a GBS isolate to inoculateadult mice intranasally and found that increased hemolysinproduction was associated with a decreased 50% lethal doseand earlier time to death for a given inoculum. Lung histopa-thology was not evaluated in that study.GBS beta-hemolysin activity is inhibited by a number of

phospholipid components of surfactant (25), and Tapsall andPhillips (42) have suggested that this inhibition may yield aclue to the increased susceptibility of premature infants, whoare deficient in surfactant, to severe early-onset GBS pneumo-nia. Our observation that DPPC, the primary component ofhuman surfactant, inhibits LDH release from lung epithelialcells exposed to beta-hemolysin is consistent with this hypoth-esis. The range of DPPC concentrations used in the inhibitionexperiments (100 to 500 mg/ml) corresponds roughly to theactual increase in concentration of DPPC found in human fetalalveolar fluid through the third trimester of pregnancy to de-livery (8). In mechanically ventilated, preterm rabbits whichreceived an intratracheal challenge of GBS, treatment withexogenous surfactant appeared to reduce inflammatorychanges on histologic examination of lung tissue (20). It isnoteworthy that pulmonary involvement is uncommon in late-onset GBS disease, which more often affects term infants (3),and recent reports of invasive GBS disease affecting immuno-compromised adults (14).The invariant link between the levels of beta-hemolysin ex-

pression and pigment production among the NH and HH GBSmutants produced in this study is consistent with the in vitroobservations of others (40, 45, 46) and suggests a genetic link-age of the two phenotypes. Although it has been shown thatthe pigment is not a carrier for the hemolysin (41), we cannotexclude definitively the possibility that the pigment in someway contributes to GBS beta-hemolysin-induced cytotoxicity.In summary, we have shown a direct correlation between

GBS beta-hemolysin expression and lung epithelial cell injury,a correlation which holds across several serotypes and with anumber of mutagenesis techniques. The isogenic, single-trans-poson-insertion GBS hemolysin mutants described in thisstudy will allow future studies to (i) identify DNA sequencesflanking the transposon insertion sites which may represent thebeta-hemolysin structural gene or elements involved in its reg-ulation, (ii) attempt to isolate the beta-hemolysin by compar-ative analysis of preparations from A909 and its NH and HHvariants, (iii) investigate a possible genetic linkage betweenbeta-hemolysin and pigment production, and (iv) directly test

FIG. 6. Inhibition of GBS hemolysin-associated epithelial cell injury by thephospholipid DPPC. The graph depicts LDH release from the HH clinical isolateNCTC 10/84, the HH mutant IN40, and starch-Tween 80 hemolysin extractsfrom each strain. Increasing DPPC concentration in the medium is associatedwith decreased release of LDH and therefore diminished cellular injury.


the role of the GBS beta-hemolysin (cytolysin) in virulence inanimal models of early-onset GBS pneumonia.

ACKNOWLEDGMENTS

We are grateful to Aphakorn Nittayajarn and Mathilde Jonas forexpert technical assistance and to Jeffrey Weiser and David Pritchardfor providing GBS hemolysin mutants.This work was supported by National Institute of Health grants HD

07233 (V.N.), and AI 30068 (C.E.R.), and AI25152 (C.E.R.).

REFERENCES

1. Ablow, R. C., S. G. Driscoll, E. L. Effmann, I. Gross, C. J. Jolles, R. Uauy,and J. B. Warshaw. 1976. A comparison of early-onset group B streptococcalneonatal infection and the respiratory distress syndrome of the newborn.N. Engl. J. Med. 294:65–70.

2. Alper, M. D., and B. N. Ames. 1975. Positive selection of mutants withdeletions of the gal-chl region of the Salmonella chromosome as a screeningprocedure for mutagens that cause deletions. J. Bacteriol. 121:259–266.

3. Anthony, B. F., and D. M. Okada. 1977. The emergence of group B strep-tococci in infections of the newborn infant. Annu. Rev. Med. 28:355–369.

4. Baker, C. J., and M. S. Edwards. 1995. Group B streptococcal infections, p.980–1054. In J. Remington and J. O. Klein (ed.), Infectious diseases of thefetus and newborn infant, 4th ed. W. B. Saunders, Philadelphia.

5. Bhakdi, S., and J. Tranum-Jensen. 1991. Alpha-toxin of Staphylococcusaureus. Microbiol. Rev. 55:733–751.

6. Bhakdi, S., J. Tranum-Jensen, and A. Sziegoleit. 1985. Mechanism of mem-brane damage by streptolysin-O. Infect. Immun. 47:52–60.

7. Chi, E. Y., T. Mehl, D. Nunn, and S. Lory. 1991. Interaction of Pseudomonasaeruginosa with A549 pneumocyte cells. Infect. Immun. 59:822–828.

8. Clements, J. A., and W. H. Tooley. 1977. Surface-active material in fetal lung,p. 349–366. In W. A. Hodson (ed.), Development of the lung. Marcel Dek-ker, Inc., New York.

9. Cochi, S. L., and R. A. Feldman. 1983. Estimating national incidence ofgroup B streptococcal disease: the effect of adjusting for birth weight. Pedi-atr. Infect. Dis. J. 2:414–415.

10. Conrads, G., A. Podbielski, and R. Lutticken. 1991. Molecular cloning andnucleotide sequence of the group B streptococcal hemolysin. Zentralbl.Bakteriol. 275:179–194.

11. Dal, M.-C., and H. Monteil. 1983. Hemolysin produced by group B strepto-coccus agalactiae. FEMS Microbiol. Lett. 16:89–94.

12. Ermert, L., S. Rousseau, H. Schutte, R. G. Birkemeyer, F. Grimminger, S.Bhakdi, H. R. Duncker, and W. Seeger. 1992. Induction of severe vascularleakage by low doses of Escherichia coli hemolysin in perfused rabbit lungs.Lab. Invest. 66:362–369.

13. Facklam, R. R., J. F. Padula, L. G. Thacker, E. C. Wortham, and B. J.Sconyers. 1974. Presumptive identification of groups A, B, and D strepto-cocci. Appl. Microbiol. 27:107–113.

14. Farley, M. M., R. C. Harvey, T. Stull, J. D. Smith, A. Schuchart, J. S.Wenger, and D. S. Stephens. 1993. A population-based assessment of inva-sive disease due to group B Streptococcus in pregnant and nonpregnantadults. N. Engl. J. Med. 328:1807–1811.

15. Framson, P. E., A. Nittayajarn, P. Youngman, and C. E. Rubens. Delivery ofTn917 to the group B streptococcus genome via pTV1OK, a temperature-sensitive pWV01 plasmid. Submitted for publication.

16. Gibson, R. L., M. K. Lee, C. Soderland, E. Y. Chi, and C. E. Rubens. 1993.Group B streptococci invade endothelial cells: type III capsular polysaccha-ride attenuates invasion. Infect. Immun. 61:478–485.

17. Gray, B. M., and D. G. Pritchard. 1992. Phase variation in the pathogenesisof group B streptococcal infections. Zentralbl. Bakteriol. Suppl. 22:452–454.

18. Griffiths, B. B., and H. Rhee. 1992. Effects of haemolysins of groups A andB streptococci on cardiovascular system. Microbios 69:17–27.

19. Hemming, V. G., D. W. McCloskey, and H. R. Hill. 1976. Pneumonia in theneonate associated with group B streptococcal septicemia. Am. J. Dis. Child.130:1231–1233.

20. Herting, E., C. Jarstrand, O. Rasool, T. Curstedt, B. Sun, and B. Robertson.1994. Experimental neonatal group B streptococcal pneumonia: effect of amodified porcine surfactant on bacterial proliferation in ventilated near-term rabbits. Pediatr. Res. 36:784–791.

21. Hull, R. A., R. E. Gill, P. Hus, B. H. Minshew, and S. Falkow. 1981. Con-struction and expression of recombinant plasmids encoding type 1 or D-mannose-resistant pili from a urinary tract infection Escherichia coli isolate.Infect. Immun. 45:528–530.

22. Jett, B. D., H. G. Jensen, R. E. Nordquist, and M. S. Gilmore. 1992. Con-tribution of the pAD1-encoded cytolysin to the severity of experimental

Enterococcus faecalis endophthalmitis. Infect. Immun. 60:2445–2452.23. Kuypers, J. M., L. M. Heggen, and C. E. Rubens. 1989. Molecular analysis of

a region of the group B streptococcus chromosome involved in type IIIcapsule expression. Infect. Immun. 57:3058–3065.

24. Madoff, L. C., J. L. Michel, and D. L. Kasper. 1991. A monoclonal antibodyidentifies a protective C-protein alpha-antigen epitope in group B strepto-cocci. Infect. Immun. 59:204–210.

25. Marchlewicz, B. A., and J. L. Duncan. 1980. Properties of a hemolysinproduced by group B streptococci. Infect. Immun. 30:805–813.

26. Marchlewicz, B. A., and J. L. Duncan. 1981. Lysis of erythrocytes by ahemolysin produced by a group B Streptococcus sp. Infect. Immun. 34:787–794.

27. Merritt, K., and N. J. Jacobs. 1976. Improved medium for detecting pigmentproduction by group B streptococci. J. Clin. Microbiol. 4:379–380.

28. Noble, M. A., J. M. Bent, and A. B. West. 1983. Detection and identificationof group B streptococci by use of pigment production. J. Clin. Pathol.36:350–352.

29. Papadimitrious, J. C., C. B. Drachenberg, M. L. Shin, and B. F. Trump.1994. Ultrastructural studies of complement mediated cell death: a biologicalreaction model to plasma membrane injury. Virchows Arch. 424:677–685.

30. Platt, M. W. 1995. In vivo hemolytic activity of group B streptococcus isdependent on erythrocyte-bacteria contact and independent of a carriermolecule. Curr. Microbiol. 31:5–9.

31. Rooney, S. A. 1985. The surfactant system and lung phospholipid biochem-istry. Am. Rev. Respir. Dis. 131:439–460.

32. Rubens, C. E., and L. M. Heggen. 1988. Tn916DE: a Tn916 derivativeexpressing erythromycin resistance. Plasmid 20:137–142.

33. Rubens, C. E., H. V. Raff, J. C. Jackson, E. Y. Chi, J. T. Bielitzki, and S. L.Hillier. 1991. Pathophysiology and histopathology of group B streptococcalsepsis in Macaca nemestrina primates induced after intraamniotic inocula-tion: evidence for bacterial cellular invasion. J. Infect. Dis. 164:320–330.

34. Rubens, C. E., S. Smith, M. Hulse, E. Y. Chi, and G. van Belle. 1992.Respiratory epithelial cell invasion by group B streptococci. Infect. Immun.60:5157–5163.

35. Rubens, C. E., M. R. Wessels, L. M. Heggen, and D. L. Kasper. 1987.Transposon mutagenesis of type III group B Streptococcus: correlation ofcapsule expression with virulence. Proc. Natl. Acad. Sci. USA 84:7208–7212.

36. Schwartz, D. O., and J. R. Beckwith. 1969. Mutagens which cause deletionsin Escherichia coli. Genetics 61:371–376.

37. Seeger, W., R. G. Birkemeyer, L. Ermert, N. Suttorp, S. Bhakdi, and H. R.Duncker. 1990. Staphylococcal alpha-toxin-induced vascular leakage in iso-lated perfused rabbit lungs. Lab. Invest. 63:341–349.

38. Southern, E. 1975. Detection of specific sequences among DNA fragmentsseparated by gel electrophoresis. J. Mol. Biol. 98:503–517.

39. Tacson, R. I., J. A. Vazquez-Boland, C. B. Gutierrez-Martin, I. Rodriguez-Barbosa, and E. F. Rodriguez-Ferri. 1994. The RTX haemolysins Apx1 andApxII are major virulence factors of the swine pathogen Actinobacilluspleuropneumonia: evidence from mutational analysis. Mol. Microbiol. 14:207–216.

40. Tapsall, J. W. 1986. Pigment production by Lancefield-group-B streptococci(Streptococcus agalactiae) J. Med. Microbiol. 21:75–81.

41. Tapsall, J. W. 1987. Relationship between pigment production and haemo-lysin formation by Lancefield group B streptococci. J. Med. Microbiol. 24:83–87.

42. Tapsall, J. W., and E. A. Phillips. 1991. The hemolytic and cytolytic activityof group B streptococcal hemolysin and its possible role in early onset groupB streptococcal disease. Pathology 23:139–144.

43. Thelestam, M., and L. Blomqvist. 1988. Staphylococcal alpha toxin—recentadvances. Toxicon 26:51–65.

44. Tsaihong, J. C., and D. E. Wennerstrom. 1983. Effect of carrier molecules onproduction and properties of extracellular hemolysin produced by Strepto-coccus agalactiae. Curr. Microbiol. 9:333–338.

45. Weiser, J. N., and C. E. Rubens. 1987. Transposon mutagenesis of group Bstreptococcus beta-hemolysin biosynthesis. Infect. Immun. 55:2314–2316.

46. Wennerstrom, D. E., J. C. Tsaihong, and J. T. Crawford. 1985. Evaluation ofthe role of hemolysin and pigment in the pathogenesis of early onset groupB streptococcal infection, p. 155–156. In Y. Kimura, S. Kotami, and Y.Shiokawa (ed.), Recent advances in streptococci and streptococcal diseases.Reedbooks, Bracknell, United Kingdom.

47. Wessels, M. R., V.-J. Benedi, D. L. Kasper, L. M. Heggen, and C. E. Rubens.1991. Type III capsule and virulence of group B streptococci, p. 219–223. InG. M. Dunny, P. P. Cleary, and L. L. McKay (ed.), Genetics and molecularbiology of streptococci, lactococci, and enterococci. American Society forMicrobiology, Washington, D.C.

48. Wilkinson, H. W. 1977. Nontypable group B streptococci isolated fromhuman sources. J. Clin. Microbiol. 6:183–184.

Editor: V. A. Fischetti


Documents

Group B Streptococcal Beta-Hemolysin Expression Is Associated